Independent variables

4.4.2.1 Industry specialisation variables

Prior research has used various measures of industry specialisation. Most of these measures are based on audit firm market share within a particular industry (see, for example, studies by Craswell et al., 1995; Ferguson and Stokes, 2002; Krishnan, 2003;

Casterella et al., 2004; Chen, Moroney and Houghton, 2005). The underlying reasoning is that the firms with the largest market shares have developed the largest knowledge base within that particular industry and have made significant investments in developing industry-specific audit technologies (Neal and Riley, 2004). We classify auditors as industry specialists if they have a within-industry market-share of at least 25 percent. Prior to the consolidation of the Big 8 into the Big 6 in 1989, the auditor specialisation literature designated auditors as industry specialists if they audited more than 10 percent of firms in the industry (see, for example, Palmrose, 1986; Defond, 1992; Craswell et al., 1995).

After the consolidation, most auditor specialisation studies used a specialisation measure of 20 percent market share (see, for example, Neal and Riley, 2004; Dunn and Mayhew, 2004; Casterella et al., 2004; Chen et al., 2005). Because the data of this study are drawn from the years 1998-2001, with all observations after the 1998 PricewaterhouseCoopers merger, a more restrictive specialisation measure seems appropriate. We set our measure

of industry specialisation at a 25 percent market share cut-off, which requires industry specialists to service a market share at least 25 percent greater than if the firms were to split the industry evenly among them. The variable SPECIALIST is set equal to one if the company is audited by a Big 5 auditor with at least 25 percent market share in the industry, and zero otherwise. The auditor's industry share (using the square root of client sales as the base) in each two-digit SIC code is computed using the population of all available observations (comprising Big 5 and non-Big 5 clients) from Compustat for the period 1998-2001¹⁰.

4.4.2.2 Audit methodology variable

Lemon, Tatum and Turley (2000) conducted the first major international study of the business risk audit methodology. This study involved a survey and detailed interviews with partners from Big 5 audit firms and some second tier firms in the UK, US, and Canada. The results from this study indicated that substantial differences exist between audit firms with respect to the implementation of this audit methodology. Due to confidentiality agreements, however, the nature and extent of these variations were not explored in this study. This issue received further attention in a more recent study by Curtis and Turley (2005), providing more details with respect to the diversity in the approaches of the Big 5 firms. The interviews in this study indicated that two of the five largest audit firms adopted the business risk methodology to a great extent, while the remaining three were somewhat lagging behind. Issues that caused much debate between the participating firms were the scope of business risks to be addressed, how such risks should be linked to the financial statements, the appropriateness of relying on high level controls and the concept and implications of ‘significant risks’. In this study, we proxy for the differences between Big 5 audit firms with regard to the adoption of the business risk methodology by including a dummy variable BRA, which is coded 1 if the company is audited by one of the two Big 5 firms that implemented the business risk audit methodology to a great extent, and 0 otherwise.

10 All but two industries included in this study comprise more than 30 companies. Excluding the three

4.4.2.3 Strategic and operating variables

We measure company turnaround approach using variables that contain information regarding turnaround strategies that have been implemented during the year under audit to overcome financial difficulties. Consistent with the categorization of management turnaround initiatives in chapter 2, we investigate the impact of three categories of management initiatives that may potentially mitigate the adverse conditions affecting performance. First, we distinguish between strategic and operating turnaround initiatives. This categorization stems from the strategic literature, where the distinction between a strategic and operating turnaround approach was first introduced by Hofer (1980). Operating initiatives aim at a short-term improvement in financial performance through for example cost-cutting, asset disposal, increased marketing efforts and upgrading existing products and processes, whereas strategic initiatives aim at long-term profitability by solving external, strategic problems through for example cooperative agreements, acquisitions or the introduction of new products.

Based on prior research with respect to the performance implications of company strategy (see, for example, Stuart, 2000; King, Dalton, Daily and Covin, 2004; Mishina, Pollock and Porac, 2004), we further categorize the strategic turnaround initiatives into initiatives that are expected to have a positive impact on the company’s liquidity status within the next 12 months and strategic approaches that are only expected to have a long-term impact on performance. We classify cooperative agreements with other firms as a strategy expected to have a positive short-term impact on performance, while the introduction of new products and acquisitions of other companies are considered strategies with a long-term impact on performance.

Based on this categorization of company turnaround initiatives, we define the variable OPERATING as a discrete variable reflecting the number of operating initiatives implemented by the company during the year under audit. The variable STRATEGIC is introduced to capture the aggregate impact of all strategic management actions. We categorize these strategic actions further into STRATST and STRATLT, which reflect the number of strategic initiatives that are expected to have a short-term and long-term impact on corporate performance.

The information regarding client operating and strategic initiatives was manually collected from the relevant 10-Ks filed with the SEC, by reading these documents cover to cover and completing a strategic scorecard. With respect to operating initiatives

(OPERATING), we assessed whether the company engaged in: (1) cost-cutting activities, (2) asset disposal, (3) upgrading existing products and processes and (4) increasing marketing efforts. This provided a discrete measure for each company with values from 0 to 4, reflecting the number of operating initiatives undertaken by the client during the year under audit. This score was subsequently divided by the maximum score in the sample to obtain a measure with a range from 0 to 1. With respect to strategic initiatives (STRATEGIC), we assessed whether the company: (1) introduced new products, (2) acquired other companies, and (3) entered into cooperative agreements with other firms.

This resulted in a discrete measure with values from 0 to 3, which was subsequently divided by its maximum score in the sample. We define STRATST as a variable that captures a company’s engagement in strategic initiatives that are capable of generating a short-term impact on performance. This variable was coded 1 if the company engaged in cooperative agreements with other firms during the year under audit, and 0 otherwise.

Finally, in order to construct a measure for strategic initiatives that are likely to have only a long-term impact on performance (STRATLT), we assessed whether the company (1) acquired other companies and (2) introduced new products during the year under audit.

This provided a discrete measure with values from 0 to 2, which was subsequently divided by the maximum score in the sample.

Another category of management initiatives to mitigate adverse conditions consists of money-raising activities identified in prior audit opinion research (Behn, Kaplan and Krumwiede, 2001; Geiger and Rama, 2003). We construct a measure RAISEMONEY, by assessing whether (1) the auditee plans to borrow funds through existing bank lines of credit or other approved debt instruments, and (2) the auditee plans to issue equity through existing or committed arrangements. The coding of these money-raising activities provides a discrete measure with values from 0 to 2, which was subsequently divided by the maximum score in the sample. The information regarding client initiatives to raise money was manually collected from the relevant 10-Ks filed with the SEC.

4.4.2.4 Control variables

Based on prior research (McKeown et al., 1991; Raghunandan and Rama, 1995;

Carcello et al., 1995; Geiger and Raghunandan 2001; Gaeremynck and Willekens, 2003;

Knechel and Vanstraelen, 2004), we include a number of control variables capturing the financial condition of the firm, bankruptcy lag, and firm size. Consistent with prior

research, we measure client size using log of net sales (LNSALES). We also use the financial distress indicator developed by Zmijewski (1984) to calculate the probability of bankruptcy (ZMIJEWSKI). In addition, we classify a company as being in default (DEFAULT) if there is either payment default or technical default of loan covenants.

Finally, we include a company’s bankruptcy lag (SQBANKRUPTLAG), reflecting the square root of days from the date of the audit report to the bankruptcy date. LNSALES and the necessary data to calculate the financial distress indicator ZMIJEWSKI were collected from the WORLDSCOPE database. SQBANKRUPTLAG and DEFAULT were calculated with information from the WORLDSCOPE database and the company’s 10-K.

The definition of the test and control variables is given in table 4.1.

T^ABLE 4.1:VARIABLE DEFINITIONS AND EXPECTED SIGNS

Variable Definition Expected

sign Dependent variable

ERROR 1 if no going-concern report was issued for a company that went bankrupt the subsequent year; 0 otherwise

Independent variables Audit Methodology

BRA 1 if the company is audited by a Big 5 auditor who adopted the

business risk methodology, 0 otherwise +/- Auditor Specialisation

SPECIALIST 1 if the company is audited by a Big 5 auditor who holds more than 25% market share (measured in square root of client net sales) in a two-digit industry, 0 otherwise

-

Management initiatives

OPERATING A score from 0 to 4, scaled by its maximum value in the sample, representing the sum of all operating initiatives (marketing, asset disposal, upgrading of products and processes, cost-cutting)

?

STRATEGIC A score from 0 to 3, scaled by its maximum value in the sample, representing the sum of all strategic initiatives (new products, acquisitions, cooperative agreements)

?

STRATST Dummy variable which equals one if the company undertakes strategic initiatives with a short-term impact (cooperative agreements)

?

STRATLT A score from 0 to 2, scaled by its maximum value in the sample, representing the sum of all strategic initiatives with a long-term impact (new products, acquisitions)

?

Money-raising activities

RAISEMONEY A score from 0 to 2, scaled by its maximum value in the sample, representing the sum of financial initiatives to raise money through the issuance of stock or additional borrowings

?

TABLE 4.1:VARIABLE DEFINITIONS AND EXPECTED SIGNS

Variable Definition Expected

sign Control variables

ZMIJEWSKI probability of bankruptcy, calculated from the Zmijewski (1984)

weighted probit bankruptcy prediction model - DEFAULT 1 if in payment default or technical default of loan covenants, 0

otherwise

- SQBANKRUPTLAG the square root of the number of days from the audit report date to

the date of bankruptcy +

LNSALES natural log of net sales ?